J Chem Ecol (2009) 35:1222–1233 DOI 10.1007/s10886-009-9705-6
Spatial Displacement of Release Point can Enhance Activity of an Attractant Pheromone Synergist of a Bark Beetle Brian T. Sullivan & Kenji Mori
Received: 3 December 2008 / Revised: 7 October 2009 / Accepted: 14 October 2009 / Published online: 10 November 2009 # US Government 2009
Abstract Flight responses of the southern pine beetle, Dendroctonus frontalis Zimmermann, to widely-spaced (>130 m) traps baited with pine volatiles (in turpentine) and the female-produced pheromone component frontalin were enhanced when a bait containing the male pheromone component (+)-endo-brevicomin was attached directly to the trap. However, displacing this bait 4–16 m horizontally from the trap significantly increased its synergistic effect. (+)-endo-Brevicomin enhanced catch to the same degree when the bait was positioned either on the trap or 32 m away. In another experiment, pairs of frontalin/turpentinebaited traps were established with 4 m spacing between traps and >100 m spacing between pairs. Attachment of either a racemic or (+)-endo-brevicomin bait to one trap of a pair caused a significant increase in catch by both traps, but catch in the trap lacking endo-brevicomin was increased more than in its endo-brevicomin-baited twin. In a third experiment, widely-spaced groups of three traps (in a line with 1 and 4 m spacing between the middle and outer traps) were baited uniformly with frontalin and turpentine, and the release rate of (+)-endo-brevicomin from the middle trap was varied across three orders of magnitude. Release rates sufficient to enhance total D. frontalis catch by the trio also caused relatively higher catches to occur in the outer traps than in the middle one. These experiments indicated that
B. T. Sullivan (*) USDA Forest Service, Southern Research Station, 2500 Shreveport Hwy, Pineville, LA 71360, USA e-mail:
[email protected] K. Mori The University of Tokyo, Bunkyo-ku, Tokyo 113-0023, Japan
both male and female D. frontalis fly to and land preferentially at sources of frontalin and host odors when these are located some distance away from a source of endo-brevicomin. This behavior may have evolved in D. frontalis to allow host-seeking beetles to locate growing, multi-tree infestations while avoiding fully-colonized trees within these infestations. Our data demonstrate that trap spacing alone can qualitatively change the outcome of bait evaluation trials and may explain why many earlier experiments with endo-brevicomin failed to identify it as an aggregation pheromone synergist for D. frontalis. We believe that important aggregative functions of semiochemicals of other bark beetle species may have been similarly overlooked due to choice of experimental procedures. Keywords Aggregation . endo-Brevicomin . Dendroctonus frontalis . Flight behavior . Host location . Pheromone synergist . Scolytidae . Semiochemical
Introduction In studies used to identify pheromone synergists and inhibitors for bark beetles (Coleoptera: Scolytidae), candidate behavioral chemicals are typically released from the same location as an attractant, and the ability of the compounds to modify insect flight to the attractant is then measured, usually with a trap (Reeve and Strom 2004; Fettig et al. 2006). However, in natural bark beetle attacks, release points of pheromones (i.e., the individual beetle gallery entrances) are dispersed in space and time. They occur along the length of the tree stem (bole) and among adjacent, attacked trees. Furthermore, the number of new galleries accumulates with time and the composition of the pheromone blend emitted from individual beetle attacks
J Chem Ecol (2009) 35:1222–1233
changes as host colonization proceeds (Birgersson and Bergström 1989). Both males and females may contribute unique components to the pheromone blend, whereas only one sex initiates colonization (Borden 1982); hence pheromone components from the pioneer sex are emitted prior to those of the joining sex, but as multiple families accumulate at an aggregation, signaling by the pioneering and joining sexes overlaps. Additionally, either sex may produce a succession of different compounds during host colonization (Birgersson et al. 1984; Byers et al. 1984). Attacks on individual trees begin frequently at a specific height on the bole and then spread to adjacent portions over the course of a few days, and adjacent trees within infestations are typically attacked successively rather than simultaneously (Gara and Coster 1968; Berryman 1982; Schlyter et al. 1987a). Hence, at any one time, the composition and release rate of the emitted pheromone blend will vary spatially, both along the length of the infested bole as well as among adjacent, attacked trees. Pheromones arising from adjacent rather than identical points in space may send a distinctly different message to host-seeking beetles, and thus they might differently influence the location where a beetle ultimately lands. However, few studies of bark beetles have examined the effect of separating the release points of aggregation pheromone components (Byers 1987). The southern pine beetle, Dendroctonus frontalis Zimmermann (Coleoptera: Scolytidae), is an aggressive bark beetle that is native to the southeastern United States, Arizona, Mexico, and parts of Central America (Payne 1980). Infestations typically occur across the landscape in spatially discrete, rapidly expanding “spots” in which adjacent trees are attacked in succession by beetles arriving from the surrounding forest as well as by re-emerging parent and brood adults from within the infestation (Gara 1967; Gara and Coster 1968). Aggregation is mediated by a complex of semiochemicals: initially-attacking females release the attractive pheromone component frontalin, which is enhanced synergistically both by monoterpene kairomones (particularly α-pinene) released by the host pine and (+)-endo-brevicomin from secondarily-arriving males (Smith et al. 1993; Sullivan et al. 2007). Additionally, the female-produced pheromone component transverbenol may substitute for host monoterpenes when the latter are at low levels, and may play an important role in initiating host colonization (Renwick and Vité 1969; Payne et al. 1978a). Dendroctonus frontalis also produces several compounds (such as verbenone and myrtenol) that inhibit response to attractants, and one or more of these may function as antiaggregation pheromones for this species (Smith et al. 1993; Sullivan 2005). There are divergent data sets on the behavioral activity of endo-brevicomin with D. frontalis. Studies performed with
1223
racemic endo-brevicomin indicated that this compound acts as an attractant antagonist (Vité and Renwick 1971; Payne et al. 1978a; Salom et al. 1992). In contrast, baits composed of pure (+)-endo-brevicomin (only enantiomer produced by the species) strongly enhanced attraction to combinations of frontalin and host odors (Vité et al. 1985; Sullivan et al. 2007), and led these authors to conclude that this compound was a component of the D. frontalis aggregation pheromone. (+)-endo-Brevicomin alone does not attract D. frontalis into traps (Sullivan et al. 2007). Observations during trapping trials with D. frontalis suggested that trap spacing alone might determine whether the addition of (+)-endo-brevicomin increased, reduced, or did not alter catch in attractant-baited traps relative to attractant-only control traps (authors’ unpublished data). We hypothesized that interacting pheromone components of bark beetles (i.e., synergists, antagonists) might convey unique information and elicit different behaviors when their sources are separated by a short distance (i.e., a few meters) rather than being either collocated or widely separated. We therefore performed tests to determine if point sources of endo-brevicomin could influence catch in distant traps and whether varying the distance between such point sources and attractant-baited traps could reverse the apparent activity of this pheromone component. We also performed initial studies to elucidate interactions of endo-brevicomin release rate with trap spacing and contrast the spatial dynamics of endo-brevicomin to other aggregationmediating semiochemicals for D. frontalis.
Methods and Materials Location, General Design, and Materials Trapping experiments were performed by using 12-unit funnel traps (Chemtica International, San Jose, Costa Rica) placed in mature, mixed pine/hardwood stands of the Homochitto National Forest in southwestern Mississippi (31.43° N, 91.19° W). The pine component was dominated by loblolly pine, Pinus taeda L., interspersed with shortleaf pine, Pinus echinata Mill. Adjacent pine forests were experiencing moderate levels of D. frontalis infestation, but according to aerial surveys no multiple-tree beetle infestations were within 1 km of the trap locations. Traps were positioned >10 m from the nearest pine and suspended (with the bottom of the trap cup 1–1.5 m above the ground) from vertical standards consisting of 1.7 cm dia. pieces of electrical conduit staked into the ground. Trap cups contained several centimeters of aqueous propylene glycol to preserve captured insects, and catch was collected at intervals of 6–18 d (longer intervals were used during periods when flying beetles were less abundant). Frontalin baits consisted of a pair of capped 400µl-capacity LDPE
1224
J Chem Ecol (2009) 35:1222–1233
microcentrifuge tubes each loaded with 200–300 μl racemic frontalin [>95% chemical purity (contaminants with no known behavioral activity), Chemtica International]. Turpentine baits consisted of a single 250 ml-capacity brown glass bottle with a piece of 1 cm dia. cotton dental wick immersed in the liquid (200–250 ml, steam distilled from P. taeda; Hercules Inc., Brunswick, GA, USA) and protruding 2.5 cm through the cap. endo-Brevicomin was released either from glass capillaries with one heat-sealed end or from open, 100 μl-capacity glass autosampler vial inserts; these were secured open-end-up inside of an uncapped, inverted 4 or 8 ml-capacity glass screw-cap vial (Table 1). The capillaries and autosampler vial inserts were secured in the vial with a silicone GC septum that was pressed sideways into the vial mouth; the sealed end of each capillary (or the tapered tip of each insert) was inserted into a pinhole in the interior-facing edge of the septum. (+)endo-Brevicomin was synthesized as described elsewhere (Sullivan et al. 2007) and was >99% enantiomerically and 95% chemically pure by GC (contaminants with no known behavioral activity). Racemic endo-brevicomin [PheroTech (now ConTech Inc.), Delta, BC, Canada] was 95% chemically pure (130 m apart, baited uniformly with frontalin and turpentine, and assigned one of six experimental treatments (Fig. 1a): A single (+)-endo-brevicomin bait was either attached directly to the trap (i.e., at 0 m), or positioned 4, 8, 16, or 32 m away, or no endo-brevicomin bait was assigned to the trap (in this case the closest source of endobrevicomin was associated with adjacent traps and thus was >100 m away). Three lines of six traps each were established, and the six treatments were assigned initially at random to the six traps of each line and then re-randomized without replacement to any previous position for each of six successive trap collections. Thus, the experimental design was three complete Latin squares with each square consisting of six traps (columns) and six successive trapping intervals (rows). The displaced endo-brevicomin release devices were suspended 1.5 m above the ground on plastic garden stakes and the direction that each release device was located relative to its treatment trap (0°, 60°, 120°, 180°, 240°, or 300° from north) was assigned at random and without duplication among the six traps of each line. This directional assignment remained the same for each trap position for the duration of the experiment. The experiment was conducted 8 February–12 April 2006. Trap catches of D. frontalis were cube root transformed and analyzed with a mixed-model ANOVA in which treatment, date, and treatment by date effects were considered fixed, whereas square, trap within square, and treatment by square effects were regarded as random (SAS 9.0, SAS Institute Inc., Cary, NC). The cube-root transformation was used for analysis of this and all following ANOVAs of total D. frontalis catches because it was generally better at normalizing residuals than the weaker square root or the stronger log10 transformation (based on examination of residual plots), and in all comparisons the residuals from the transformed data sets did not fail the Kolmogorov-Smirnov test for a normal distribution (α=0.05). Treatment means
Table 1 Construction and elution rate of endo-brevicomin release devicesa Glass insert/capillary characteristics Experiment number (release rate) 1, 2b, 4 3 (low rate) 3 (medium rate) 3 (high rate)
Diameter (i.d.) 1.2 0.6 1.2 3.7
mm mm mm mm
Length 20 32 32 30
mm mm mm mm
Height filled 10 10 10 8
mm mm mm mm
Number per device
Elution ratec (mg/d ± s.d.)
1 1 3 3
0.23±0.01 0.045±0.007 0.52±0.03 2.8±0.6
a
Each device consisted of 1–3 glass autosampler vial inserts or glass capillaries secured open-end-up inside of an inverted, uncapped 4 ml (8 ml for the high release rate device) capacity screw-capped vial.
b
In experiment 2, a single device was used for (+)-endo-brevicomin, but two devices were used for racemic endo-brevicomin.
c
Measured in a fume hood at 23±2°C (N=3).
J Chem Ecol (2009) 35:1222–1233
a Experiment 1
T
Six treatments: 0, 4, 8, 16, 32, or >100 m +B
F
b T
Three treatments:
X
X = +B, ±B, or nothing
F
T
“4 m trap”
“0 m trap”
Experiment 2
F
4m
c
F
T
Four treatments:
X
X = +B at high, medium, or low dose, or absent
“4 m trap”
T
“0 m trap”
“1 m trap”
Experiment 3
F
T
F
4m
1m
d
X
1m
T “0 m trap”
Y
+B
Three treatments: X,Y = F,T; +B,T; or F,+B
F
4m
“4 m trap”
Experiment 4
“1m trap”
Fig. 1 Trap and bait arrangements for experiments 1–4. Rectangles represent release devices of either frontalin (F), turpentine (T), (+)-endo-brevicomin (+B), racemic endo-brevicomin (±B), or an experimentally variable bait (shaded X and Y). For the 4, 8, 16, and 32 m-distances in experiment 1, the “+B” release device was attached to the top of a plastic garden stake; for the >100 m treatment, no+B device was assigned to the trap and the closest such device was associated with an adjacent treatment. Otherwise, release devices were attached directly to traps
1225
Y X
1226
were compared by using Tukey’s test for all pairwise comparisons (α=0.05) with treatment by square as the error term. In this and the other trapping experiments, catches for some traps and intervals were very low or zero, hence use of the above ANOVA for analysis of treatment effects on sex ratio was not possible due to the excessive numbers of missing values and/or the highly non-normal distributions of residuals. Therefore, the proportion of female D. frontalis trapped in each treatment was pooled within a square and subjected to a two-way ANOVA with square and treatment as factors. Proportions of females trapped were not transformed for this or other experiments since the pooled data never failed the Kolmogorov-Smirnov test (α=0.05). Experiment 2: Pairs of Adjacent Traps Differing in endoBrevicomin Pairs of funnel traps spaced 4 m apart and baited identically with frontalin and turpentine were established at locations separated by >100 m (Fig. 1b). One trap of each pair (the “0 m trap;” chosen by a coin toss), received either a single (+)-endo-brevicomin bait, two racemic endo-brevicomin baits, or no additional bait (three experimental treatments). The opposite trap (the “4 m trap”) received no additional bait. Four lines, each with three trap pair locations, were established, and the treatments were assigned initially at random to the three pairs of each line and then re-randomized without replacement to any previous pair for each of three successive trap collections. Thus, the experimental design was four complete Latin squares with each square consisting of three adjacent trap pairs (columns) and three successive trapping intervals (rows). The experiment was run 13 December 2005–17 January 2006. The summed D. frontalis catch per trap pair (Σx = X4m + X0m, transformed 3√Σx) and the difference in catch (Dx = X4m-X0m, transformed 3√Dx), were analyzed with the mixed-model ANOVA of experiment 1. Treatment effects on the proportion of females trapped per pair and the difference in proportion of females between members of each pair were analyzed with a two-way ANOVA (as in experiment 1). Additionally, beetle responses to the two traps of each pair (X4m vs. X0m) were compared within each treatment by a paired t-test (α=0.05). Experiment 3: Trios of Adjacent Traps with Varying endoBrevicomin Release Rate At locations separated by >100 m, groups of three traps were erected in a straight line with the outside traps spaced 1 m and 4 m away from the middle trap (Fig. 1c). The outside traps were erected first, and then the location of the middle trap respective to either outside trap (i.e., the assignment of 4 m and 1 m-distant traps) was chosen by a coin toss. All three traps were baited with frontalin and turpentine, and the middle trap (the “0 m trap”) either received no additional bait or was baited additionally with a device releasing either a low, medium, or high rate
J Chem Ecol (2009) 35:1222–1233
(Table 1) of (+)-endo-brevicomin (four experimental treatments). Two lines, each with four trap trio locations, were established, and the treatments were assigned initially at random to the four trios of each line and then re-randomized without replacement to any previous trio for each of four successive trap collections. After these initial four collections (after every treatment had been at every trio location once), the middle trap was moved 3 m in order to reverse the designation of the 1- and 4 m-distant traps within each trio. Then treatment assignments were re-randomized among trios within lines, and a second set of four successive collections was performed similar to the first. Thus, the experimental design was four complete Latin squares with each square consisting of four adjacent trap trios (columns) and four successive trapping intervals (rows). The experiment was run 16 August–23 October 2005. Total catch of D. frontalis per trap trio (Σx = X4m + X1m + X0m; transformed 3√Σx) as well as the differences in catch between the outside and middle traps (D1 = X4m-X0m, D2 = X1m-X0m; transformed 3√Dx) were analyzed with a mixed model ANOVA in which treatment and date within square were considered fixed, and square, trio within square, and treatment by square were regarded as random effects (SAS 9.0). Date was included in the model as a fixed effect nested within square to account for replication of squares in time. Tukey’s all-pairwise comparisons (α=0.05) utilized treatment by square as the error term. Treatment effects on the proportion of females trapped per trio, and the differences in proportion of females between the outer and middle traps, were analyzed with a two-way ANOVA (as in experiment 1). Experiment 4: Trios of Adjacent Traps with Variable Baits Trap trios were established identically as in experiment 3, however, the 0 m trap was baited consistently with frontalin, turpentine, and (+)-endo-brevicomin, whereas the 4- and 1 m-distant traps were both baited identically with just two of these bait components and thus lacked one of (+)-endo-brevicomin, frontalin, or turpentine (i.e., three treatments, Fig. 1d). Three lines, each with three trap trio locations were established, and treatments were assigned initially at random to the three trios of each line and then rerandomized without replacement to any previous position for each of three successive trap collections. Thus, the experimental design was three complete Latin squares, with each square consisting of three adjacent trap trios (columns) and three successive trapping intervals (rows). The experiment was run 17 October–14 November 2005. Differences in catch between the outside and middle traps (D1 = X4mX0m, D2 = X1m-X0m; transformed 3√Dx ) were compared among treatments by using the mixed-model ANOVA of experiment 1. Differences in proportion of females trapped between the outer and middle traps were subjected to a twoway ANOVA (as in experiment 1).
J Chem Ecol (2009) 35:1222–1233
1227
A corollary data set was collected to determine whether catch in the outside and middle traps differed when all three traps in each trio were baited identically. Trios as described above were established at 14 different sites, all three baits were placed at all three traps, and catch during a single 7– 18 d interval was collected from each on 7 November 2005–17 January 2006. For all bait configurations, the transformed differences in catch between the outside and middle traps (3√D1, 3√D2) were each tested against the null hypothesis of equality to zero with a one-sample t-test (twotailed, α=0.05).
Results Experiment 1: Displacement of (+)-endo-Brevicomin Bait from an Attractant-Baited Trap Catch of D. frontalis in traps baited with frontalin and turpentine was significantly influenced by trap distance from a (+)-endo-brevicomin release device (F = 85.8; df = 5, 10; P < 0.001). Traps caught significantly more beetles when the (+)-endobrevicomin releaser was 0–32 m distant rather than >100 m away (Fig. 2). Furthermore, catch was significantly greater when the (+)-endo-brevicomin release device was 4, 8, or 16 m away from the trap rather than attached directly to it. Distance of the endo-brevicomin releaser from the trap did not significantly alter the
Experiment 2: Pairs of Adjacent Traps Differing in endoBrevicomin Pairs of traps spaced 4 m apart and baited identically with frontalin and turpentine caught significantly more D. frontalis when one of the two traps was baited additionally with either (+)- or racemic endobrevicomin (F=94.6; df=2, 6; P0.05). Furthermore, the disparity in catch between traps within pairs was significantly altered by the addition of either (+)- or racemic endo-brevicomin to one of the traps (F= 26.0; df=2,6; P=0.001). Within endo-brevicomin-treated trap pairs, the trap that lacked endo-brevicomin caught significantly more beetles than its endo-brevicomin-baited twin [for (+)-endo-brevicomin: t=3.78, df=11, P=0.003; for racemic endo-brevicomin: t=3.2, df=11, P=0.008], whereas there was no significant difference in catch between traps within pairs when endo-brevicomin was not added to either of the two traps (t=0.86, df=11, P= 0.41) (Fig. 3). There were no significant treatment effects on the proportions of females trapped [either total catch per trap pair (F=0.38; df=2,6; P=0.70) or difference between the traps in proportion of females caught (F= 0.71; df=2,6; P=0.53)]. The overall proportion of females trapped was 0.42±0.03 (mean ± s.d.). 14
e
50
Dendroctonus frontalis / trap / day
Dendroctonus frontalis / trap / day
60
proportion (0.40±0.01, mean ± s.d.) of females responding (F=0.37; df=5, 10; P=0.86).
de
40
cd 30
b bc
20
10
a 0
0
4 8 16 32 >100 Trap distance (m) from (+)-endo-brevicomin bait
Fig. 2 Experiment 1: Mean (± s.e.) daily catch of Dendroctonus frontalis in individual multiple-funnel traps baited with frontalin and turpentine. Traps were positioned >130 m apart, and each trap either had a single release device of (+)-endo-brevicomin placed a specified distance (0, 4, 8, 16, 32 m) away from it or had no (+)-endobrevicomin bait assigned [i.e., the closest (+)-endo-brevicomin bait was in an adjacent treatment or >100 m away)]. Means associated with the same letter were not significantly different (Tukey test on cube root transformed data, α=0.05). Means and s.e.’s were calculated from the untransformed data with six observations averaged within each Latin square (N=3)
endo-Brevicomin at 0 m trap:
12
(+)-Enantiomer b* Racemic b*
10
Absent a
8 6 4 2 0
0m
4m Trap position
Fig. 3 Experiment 2: Mean (± s.e.) daily catch of Dendroctonus frontalis in traps arranged in pairs with 4 m spacing between traps and >100 m spacing among pairs. All traps were baited with frontalin and turpentine, and one randomly-chosen trap per pair (the “0 m trap”) was baited additionally with racemic endo-brevicomin, (+)-endobrevicomin, or nothing. Treatment names in legend inset associated with the same letter did not differ significantly in summed beetle catch per pair (Tukey test on cube root transformed data, α=0.05); those with an asterisk had a significant difference in catch between the two traps within the pair (paired t-test; α=0.05). Means and s.e.’s were calculated from the untransformed data with three observations averaged within each Latin square (N=4)
1228
J Chem Ecol (2009) 35:1222–1233
Experiment 3: Trios of Adjacent Traps with Varying endoBrevicomin Release Rate Total catch of D. frontalis by trios of frontalin/turpentine-baited traps was significantly affected by the rate of (+)-endo-brevicomin released from the middle trap (F=34.3; df=3,9; P
